Bugs in 25

hi joreg, hi elias, hi gregsn and everybody, this now for sure, founded a bug related to framedelay and filestream.
I m beginning it with beta25.
i m encounterring on the beginning of a new project strange behaviours, wich are related to this thread also:

Im using the framedelay technique to get back from a filestream its duration and enter the duration as Endtime when a loop on the complete file is needed.
Somehow framedelay / filestream is a trojandoor for strange things:

I have some videos ( 320*240; AVI raw) played for tracking with colortracker.
They are readen on only one player. they are not heavy at all, 100Mo max

Sometimes, when i change the file readen, its loop is done not to the right point, despite the duration showing is ok and well connected, and shows up the right duration time.

a simple question for filestream node: refreshment of the EndTime is done on change ? is their a kind of initialisation process when we change the file to read and takes the wrong time from framedelay ?
is there any better way to take the duration time ?

suggestion: “trojan backdoor” for undesired data (memory leaks ) from framedelay input ( memory allocated ?)

Other bug, this morning i discovered opening in my main patch a patch done at home on 25, that their was a weakness in 25: IO boxes diseapearing, enumerations transformed as empty boxes. IOBOX transformed as string… waoaow… I had to rewrite a big part of what was made at home to correct the disappearance of so many boxes. Next time i will not work at home… :-)

Maybe this will help to identify the messy messy due to my bad habits of work:

I m using Main loop 25front/25behind fps
Using Queue(Texture) with 60 as max spread for the queue ( little textures in 640480, png)
Using S and R (values) ( many) and S and R (Nodes)( some) for sending same texture to different quads .
No 3D, only quads job and 1 fx ( distorsion 2D).
1 main renderer, 2 or 3 little intermediate renderers , a dx9 drawing renderer
Resolution of diffusion 1440
actually the project is at its beginning and i have just a few subpatch working around , not a heavy thing, and I think that I am on a clean structure.

Voilà ! hope this will help , yours, karistouf


did you know that instead of writing the endtime, you can write 9999999, filestream will then use min(videolenght, pin input) as end time… so you dont need the framedelay…

no i didnt many thanks !

ok. setting 999999 to loop completely is resolving the Loop trouble.

For sure now, the responsible is framedelay